Postdoc: AI‑Driven Fast CT Reconstruction for Natural History

Danmarks Tekniske Universitet; Ørsted; On-site; DKK 450,000 - 650,000; Full time

Job summary

DTU Compute – Department of Mathematics and Computer Science in Denmark seeks an ambitious early-career researcher for advanced image reconstruction methods within the Natural Heritage 3D project. The role focuses on fast, automated CT reconstruction and computational pipelines for digitizing natural history specimens.

You will develop and validate ML-based and regularization methods, work with simulations and real data, and contribute to open-source cores like Core Imaging Library and qim3D.

Qualifications

• PhD in computational imaging, ideally X-ray CT reconstruction.
• Experience developing deep learning and/or regularization-based reconstruction methods.
• Experience with scientific Python programming.
• Experience handling both simulations and large real data.
• Strong spoken and written English.
• Strong collaboration skills and being self-driven.

Responsibilities

• Develop fast reconstruction methods for CT data from sparse or noisy data.
• Validate numerical implementations on simulated and real CT data.
• Compare supervised and self-supervised ML-based reconstruction with conventional methods.
• Develop computational reconstruction pipelines for a fast robotic CT system in collaboration with scientific software developers.
• Contribute to open-source software communities around Core Imaging Library and qim3D.
• Publish scientific articles with colleagues in physics and natural history.

Skills

Deep learning

Python programming

Reconstruction methods

Scientific communication

Education

PhD in computational imaging

Tools

Python libraries (NumPy/SciPy)

Job description

DTU Compute – Department of Mathematics and Computer Science in Denmark seeks an ambitious early-career researcher for advanced image reconstruction methods within the Natural Heritage 3D project. The role focuses on fast, automated CT reconstruction and computational pipelines for digitizing natural history specimens.

You will develop and validate ML-based and regularization methods, work with simulations and real data, and contribute to open-source cores like Core Imaging Library and qim3D.